Interest Point Detectors

2203 WordsJan 31, 20189 Pages
Interest Point Detectors Introduction: How it started Our original idea was to implement the beginning stages of a paper on measuring the driver fatigue detection from a sequence of images []. Majority of accidents reported are due to driver fatigue. One of the important parameter to be detected to measure the driver’s alertness is to track his/her eyes in the given image sequence. Eye detection is an active research area in computer vision. Applications can range from face detection to biometrics. Canthus (Eye corners) is a very stable feature that can be constantly detected from a facial image irrespective of the eye status or gaze direction. Canthus is initially defined and treated as a corner (interest point). Interest point detectors like Harris and SUSAN are being studied and discussed in this report. These corner detectors are evaluated to find which one suits best for this application. Interest Points An interest point is a relatively important kind of localized two-dimensional image structure with a mathematical description. It can be defined as the intersection of two edges or as a point with two dominant and different edge directions. “Corners”, "T", "Y" and "X" junctions have the same definition and fall under this category. Although the interest points constitute a very small portion in the image, the information on shape is concentrated at these points. Image corner detection is an important task in various computer vision and image understanding
Open Document